COLUMBIA GLACIER 287 previous to 1892, though not necessarily of more than a mile, on the basis of the marine shells close to the ice front, assuming that these sea forms would not thrive in such cold, milky water.1 That the glacier has not recently been as far out as in 1892 and 1910, is proved by the extension of mature forest with trees at least 75 to 100 years old, up to the very border of the ice. Peat three or four feet thick has accumulated in grassy glades on hilltops and flattish ground, while living trees grow only on rock or other drained elevations and along stream courses, dead trunks standing where overwhelmed by the thickening tundra masses. These relationships of vegetation, which require time for their establishment, are found in Prince William Sound but not in the Yakutat Bay region, where the glaciers were recently much more expanded. Into these vegetation areas the advancing ice of 1892 and 1909-10 plowed its way. Perhaps, as Gilbert suggests, the important minimum of Columbia Glacier that preceded the 1892 advance occurred in the nineteenth century, when Petroff does not mention a tidal glacier in Columbia Bay, though he does assign one to Port Valdez. As we have attempted to show in our preceding discussion of the subject, there are grave reasons for doubting the existence of a large tidal glacier in Port Valdez fiord at the time Petroff mentions. Within a year or so of 1892 Columbia Glacier had an important advance, since which there have been two periods of retreat and two of advance, the last of which was still in progress in 1911. The cause for these recent oscillations is not known. They may be climatic or they may be due to earthquake avalanching. If the latter is the cause, we should be inclined to ascribe the advances to avalanches during earthquakes in the Chugach Mountains, not to the Yakutat Bay earthquakes of September, 1899. Both climatic and seismic data are too meagre for settling this question, but in 1907-08 there was a 148 inch increase of snowfall2 at Valdez, 25 miles to the east, (from 299 inches in 1906-07 to 447 inches in 1907-08). The 1906-07 snowfall at Valdez seems to be below the normal amount. It decreased to 187 inches in 1904-05. There was a 887 inch increase in snowfall at Valdez (Fort Liscum) during the winter of 1901-02 (see table at end of Chapter XV). On the other hand, there were severe earthquakes in tjijs part of the Chugach Mountains in May, 1896, August, 1898, October, 1900, March, 1903, and February, 1908. Since the 1892 advance, whose cause we will not discuss, the first of the recent advances came between 1899 and 1905, and may have been due to the 1896 or 1898 or 1900 earthquakes. The advance which was in progress at the time of our visits in 1909 and 1910 probably began in 1907 and 1908 and may have been caused by one of the preceding earthquakes, or, as already suggested, it may be that both advances were caused by variations of snowfall or of temperature. In the latter case other adjacent glaciers should have advanced and retreated, for notable temperature variation can hardly be as localized as precipitation may be. In view of the limited data at hand it is not profitable further to discuss the cause of the fluctuations of Columbia Glacier, though we are convinced that it is well worth while i Grant, U. S. and EGggins, D. P., Glaciers of the Northern Port of Prince William Sound, Bull. Amer. Geog. Soc., Vol. XLH, 1910, pp. 729-^TSl.. i Martin, Lawrence, Zeitachrift fttr Gletscherkunde, Band VIE, 1913, pp. 28-81.